Post by Julian Harris

Claude Mythos leak: “dramatically better than Opus 4.6”. “Very expensive to serve”. Restricted initially to cybersecurity partners. Leaked blog post below "Introducing Claude Mythos "Mythos" is a new name for a new tier of model: larger and more intelligent than our Opus models — which were, until now, our most powerful. We chose the name to evoke the deep connective tissue that links together knowledge and ideas. "Compared to our previous best model, Claude Opus 4.6, Mythos gets dramatically higher scores on tests of software coding, academic reasoning, and cybersecurity, among others. "In preparing to release Claude Mythos, we want to act with extra caution and understand the risks it poses — even beyond what we learn in our own testing. In particular, we want to understand the model's potential near-term risks in the realm of cybersecurity — and share the results to help cyber defenders prepare. "Mythos is also a large, compute-intensive model. It's very expensive for us to serve, and will be very expensive for our customers to use. We're working to make the model much more efficient before any general release. "For those reasons, we're taking a slower, more gradual approach to releasing Mythos than we have with our other models. We're beginning with a small number of early-access customers, who will explore the model's cybersecurity applications and report back what they find. A head start for cybersecurity We have written several times in recent months about the rapid progress in AI models' cybersecurity skills — skills that can be used for good or for ill. We've documented the ways in which models can be used to rapidly discover vulnerabilities in codebases; we've also shown how they're already being used to commit large-scale cyberattacks. Although Mythos is currently far ahead of any other AI model in cyber capabilities, it presages an upcoming wave of models that can exploit vulnerabilities in ways that far outpace the efforts of defenders. That's why our release plan for Mythos focuses on cyber defenders: we're releasing it in early access to organizations, giving them a head start in improving the robustness of their codebases against the impending wave of AI-driven exploits. Pre-release safety testing As with all of our models, we have tested Claude Mythos on a very wide variety of safety and capability evaluations. Expanding the release We'll be slowly expanding access to Claude Mythos to more customers using the Claude API over the coming weeks. Since we're particularly interested in cybersecurity uses, that's where we aim to expand the EAP initially."